US7724940B2ExpiredUtilityPatentIndex 32
Method of accurately determining similarity between target image and reference image
Est. expirySep 30, 2025(expired)· nominal 20-yr term from priority
Inventors:PARK YUN-WON
G06V 10/7515H04N 19/107
32
PatentIndex Score
0
Cited by
6
References
14
Claims
Abstract
Provided is a method of determining a similarity between a target image and a reference image. The method includes (a) obtaining related approximation pixel-values of the reference image p t having overall relationship between coordinates (x, y) and original pixel-values f t of pixels of the reference image; (b) obtaining related approximation pixel-values of the target image p s having overall relationship between coordinates (x, y) and original pixel-values f s of pixels of the target image; and (c) determining the similarity using the target related approximation pixel-values p s and the reference related approximation pixel-values p t .
Claims
exact text as granted — not AI-modified1. A method of determining a similarity between a target image and a reference image using a vision system included in a part mounter, the method comprising:
(a) obtaining related approximation pixel-values of the reference image, wherein the related approximation pixel-values have an overall relationship between coordinates and original pixel-values f t of pixels of the reference image;
(b) obtaining target related approximation pixel-values having overall relationship between coordinates and original pixel-values of pixels of the target image; and
(c) determining the similarity using the target related approximation pixel-values and the reference related approximation pixel-values,
wherein operation (b) comprises:
(b1) setting a polynomial equation having variables and unknown constants corresponding to the coordinates (x, y) of the pixels of the target image;
(b2) completing the polynomial equation by obtaining the unknown constants of the polynomial equation under a condition that can minimize differences between the original pixel-values f s of the pixels of the target image and target related approximation pixel-values p s to be obtained as a result of the polynomial equation; and
(b3) obtaining the target related approximation pixel-values p s by respectively substituting values of the coordinates (x, y) for the polynomial equation completed in operation (b2).
2. The method of claim 1 , wherein algorithms for performing operations (a) and (b) are identical.
3. The method of claim 2 , wherein, in operation (b1), the unknown constants are a 10 , a 01 , a 20 and a 02 , and a quadratic polynomial equation
p s =a 00 +a 10 x+a 01 y+a 20 x 2 +a 11 xy+a 02 y 2
is set.
4. The method of claim 3 , wherein, in operation (b1), assuming that a matrix A s of the unknown constants of the quadratic polynomial equation is
A
s
=
[
a
00
a
10
a
01
a
20
a
11
a
02
]
,
a transposed matrix of the matrix A s is A s ′, and a coordinate matrix X is
X
=
[
1
x
y
x
2
x
y
y
2
]
,
the quadratic polynomial equation is converted into a determinant by p s =A s ′X.
5. The method of claim 4 , wherein, in operation (b2), the unknown constants A s of the quadratic polynomial equation are obtained under a condition that can minimize a sum E s of squares (p s −f s ) 2 of the differences between the original pixel-values f s of the pixels of the target image and the target related approximation pixel-values p s obtained as a result of p s =A s ′X.
6. The method of claim 5 , wherein, in operation (b2), assuming that a value Xf s is obtained for each of the pixels of the target image by multiplying the coordinate matrix X by the original pixel-values f s , that a sum ΣXf s of the values Xf s is a sensed concentrative moment U sf of the target image, that a coordinate-multiplication matrix obtained after the coordinate matrix X is multiplied by a transposed matrix X′ is XX′, and that an inverse matrix of the coordinate-multiplication coordinate XX′ is B −1 , the unknown constants A s of the polynomial equation are calculated using A s =B −1 U sf .
7. The method of claim 6 , wherein, in operation (c), assuming that a transposed matrix of a connection concentrative moment U sp of the target image is U′ sp and that a connection concentrative moment of the target image is U tp , the similarity h p is given by
h
p
=
U
sp
′
B
-
1
U
tp
U
sp
′
B
-
1
U
sp
.
8. A method of determining a similarity between a target image, which is an image of portion of a picked up part, and a reference image using a vision system included in a part mounter, the method comprising:
(a) obtaining related approximation pixel-values of the reference image having overall relationship between coordinates and original pixel-values of pixels of the reference image;
(b) obtaining target related approximation pixel-values having overall relationship between coordinates and perceived luminacnes of pixels of the target image; and
(c) determining the similarity using the target related approximation pixel-values and the reference related approximation pixel-values.
9. A part mounter, the part mounter comprising:
a part-mounting head;
a nozzle coupled to the part-mounting head;
a camera arranged across from the nozzle;
a vision system coupled to the camera;
a main controller coupled to a driving system and the vision system, wherein the main controller controls the driving system based upon position information received from the vision system; and wherein the vision system photographs a target part, thereby obtaining a target image, and determines a target part type by:
(a) obtaining reference related approximation pixel-values of a reference image having an overall connection between coordinates and original pixel values of the reference image;
(b) obtaining target related approximation pixel-values having overall connection between coordinates and original pixel values of pixels of the target image; and
(c) determining a similarity between the target image and the reference image using the target related approximation pixel values and the original pixel values,
wherein operation (b) comprises:
(b1) setting a polynomial equation having variables and unknown constants corresponding to the coordinates (x, y) of the pixels of the target image;
(b2) completing the polynomial equation by obtaining the unknown constants of the polynomial equation under a condition that can minimize differences between the original pixel-values f s of the pixels of the target image and target related approximation pixel-values p s to be obtained as a result of the polynomial equation; and
(b3) obtaining the target related approximation pixel-values p s by respectively substituting values of the coordinates (x, y) for the polynomial equation completed in operation (b2).
10. The part mounter of claim 9 , further comprising an illuminator coupled to the camera, wherein the illuminator comprises a plurality of light sources.
11. A method of determining a similarity between a target image and a reference image using a vision system included in a part mounter, the method comprising:
(a) obtaining related approximation pixel-values of the reference image, wherein the related approximation pixel-values have an overall relationship between coordinates and original pixel-values f t of pixels of the reference image;
(b) obtaining target related approximation pixel-values having overall relationship between coordinates and original pixel-values of pixels of the target image; and
(c) determining the similarity using the target related approximation pixel-values and the reference related approximation pixel-values;
wherein:
algorithms for performing operations (a) and (b) are identical;
operation (b) comprises:
(b1) setting a polynomial equation having variables and unknown constants corresponding to the coordinates (x, y) of the pixels of the target image;
(b2) completing the polynomial equation by obtaining the unknown constants of the polynomial equation under a condition that can minimize differences between the original pixel-values f s of the pixels of the target image and target related approximation pixel-values p s to be obtained as a result of the polynomial equation; and
(b3) obtaining the target related approximation pixel-values p s by respectively substituting values of the coordinates (x, y) for the polynomial equation completed in operation (b2);
in operation (b1), the unknown constants are a 10 , a 01 , a 20 and a 02 , and a quadratic polynomial equation
p s =a 00 +a 10 x+a 01 y+a 20 x 2 +a 11 xy+a 02 y 2
is set; and
in operation (b1), assuming that a matrix A s of the unknown constants of the quadratic polynomial equation is
A
s
=
[
a
00
a
10
a
01
a
20
a
11
a
02
]
,
a transposed matrix of the matrix A s is A s ′, and a coordinate matrix X is
X
=
[
1
x
y
x
2
xy
y
2
]
,
the quadratic polynomial equation is converted into a determinant by p s =A s ′X.
12. The method of claim 11 , wherein, in operation (b2), the unknown constants A s of the quadratic polynomial equation are obtained under a condition that can minimize a sum E s of squares (p s −f s ) 2 of the differences between the original pixel-values f s of the pixels of the target image and the target related approximation pixel-values p s obtained as a result of p s =A s ′X.
13. The method of claim 12 , wherein, in operation (b2), assuming that a value Xf s is obtained for each of the pixels of the target image by multiplying the coordinate matrix X by the original pixel-values f s , that a sum ΣXf s of the values Xf s is a sensed concentrative moment U sf of the target image, that a coordinate-multiplication matrix obtained after the coordinate matrix X is multiplied by a transposed matrix X′ is XX′, and that an inverse matrix of the coordinate-multiplication coordinate XX′ is B −1 , the unknown constants A s of the polynomial equation are calculated using A s =B −1 U sf .
14. The method of claim 13 , wherein, in operation (c), assuming that a transposed matrix of a connection concentrative moment U sp of the target image is U′ sp and that a connection concentrative moment of the target image is U tp , the similarity h p is given by
h
p
=
U
sp
′
B
-
1
U
tp
U
sp
′
B
-
1
U
sp
.Cited by (0)
No later patents cite this yet.
References (0)
No backward citations on record.